Evidence that AGL17 is a significant downstream target of CLF in floral transition control

ABSTRACT

Polycomb-group (PcG) proteins are evolutionarily conserved in higher organisms and play essential roles in many developmental processes by catalyzing the trimethylation of histone H3 lysine 27 (H3K27me3), a key repressive histone mark. In Arabidopsis (Arabidopsis thaliana), histone methyltransferase CURLY LEAF (CLF) is one of the major PcG catalytic components, playing critical roles in plant growth and development, especially the floral transition. We have recently profiled the genome-wide occupancy of CLF by chromatin immunoprecipitation followed by high-throughput sequencing (ChIP-seq). Interestingly, AGAMOUS-LIKE 17 (AGL17) that encodes a known flowering activator was found to be a CLF direct target. Based on this observation, we set out to examine to what extent this genetic module regulates the flowering. First, we validated the ChIP-seq results by ChIP-qPCR and show that CLF indeed targets AGL17, and the level of H3K27me3 is decreased when CLF is lost. Further, we show that the expression of AGL17 is significantly up-regulated in the clf-29 mutant compared to wild-type (WT). Finally, our clf agl17 double mutant analysis suggests that AGL17 is a significant downstream target of CLF in floral transition control.

Polycomb-group (PcG) proteins mainly form two protein complexes, Polycomb repressive complex 1 (PRC1) and PRC2, which are important epigenetic regulators in plants and animals. PRC2 represses gene expression by catalyzing the trimethylation of histone H3 lysine 27 (H3K27me3).¹⁻⁶ In Arabidopsis (Arabidopsis thaliana), CURLY LEAF (CLF) is one major component of the PRC2 complex that can catalyze H3K27me3, playing important roles in plant growth and development.^7,8 Loss of CLF activity leads to plants showing severe defects such as small stature, curling leaves, and early flowering compared to wild-type (WT) plants.^9-12

We have recently profiled the genome-wide occupancy of CLF by chromatin immunoprecipitation followed by high-throughput sequencing (ChIP-seq).¹² One main focus of that work was attempted to answer how CLF modulates floral transition. Indeed, the data provided some clues for the early flowering phenotype of clf mutants. From the list of genes that are targeted by CLF, AGAMOUS-LIKE 17 (AGL17) drew our attention. AGL17, a MADS-box gene, was reported to be preferentially expressed in roots.¹³ A later study reported that the AGL17 transcript can be detected in various plant tissues with the highest expression level in roots.¹⁴ Loss-of-function of AGL17 delays flowering, while over-expression of AGL17 promotes flowering via the up-regulation of the floral meristem identity genes APETALA 1 (AP1) and LEAFY (LFY).¹⁴ Based on these published observations, we hypothesize that CLF may regulate flowering by directly targeting, and thereby repressing, the flowering activator AGL17.

As shown in Figure 1a, our ChIP-seq data showed clearly that CLF targets AGL17 directly.¹² To confirm the occupancy of CLF at AGL17, a ChIP-qPCR experiment was performed to examine the CLF enrichment at AGL17. The results showed that CLF is significantly enriched over the first exon of AGL17 (Figure 1b). In addition, it was also shown in our ChIP-seq data that the level of H3K27me3 was decreased at AGL17 in clf-29 compared to WT (Figure 1a;¹²). We validated the finding by a ChIP-qPCR experiment: as shown in Figure 1c, the H3K27me3 level was significantly decreased in clf-29 compared to WT. Further, a qRT-PCR experiment was conducted to examine the expression of AGL17 in two-week-old WT and clf-29 seedlings. We found that the expression of AGL17 was significantly up-regulated in clf-29 compared to WT (Figure 1d). Taken together, our new results clearly show that CLF directly targets AGL17 to repress its expression in Arabidopsis seedlings.

Figure 1.

CLF directly targeting AGL17 to repress its expression and control flowering time. (a) Screenshot of CLF occupancy, and H3K27me3 in WT and clf-29 at the AGL17 locus in genome browser [ENPG, www.plantseq.org;¹²] to visualize peaks. Y-axis represents the means of normalized reads (1× sequencing depth normalization) per 10 bp non-overlapping bins. Gene structure is shown underneath the panel. P1, P2, and P3 indicate the location of three pairs of primers used in (b) and (c). (b) ChIP-qPCR validation of the CLF occupancy at AGL17 by using ChIP DNAs extracted from two-week-old clf-29 pCLF::CLF-GFP seedlings. TA3, a transposable element gene, was used as a negative control locus. Error bars show standard deviations among three biological replicates. Student’s t-test, **P < .01. (c) ChIP-qPCR analysis of H3K27me3 and H3 levels in WT and clf-29 two-week-old seedlings. ACTIN 7 (ACT7) was used as a negative control. Error bars show standard deviations among three biological replicates. Student’s t-test, **P < .01. (d) Expression of AGL17 in two-week-old WT and clf-29 seedlings. The expression levels were normalized to that of GAPDH. Error bars show standard deviations among three biological replicates. Student’s t-test, **P < .01. (e) Expression of AGL17 in two-week-old WT and agl17-3 seedlings. The expression levels were normalized to that of GAPDH. Error bars show standard deviations among three biological replicates. Student’s t-test, **P < .01. (f) Image of 5-week-old WT, clf-29, agl17-3, and clf-29 agl17-3 plants grown under long-day conditions. Scale bar: 1 cm. (g) Rosette leaf number at bolting of plants in different genetic backgrounds (WT, clf-29, and clf-29 agl17-3) grown under long-day conditions. Red dots represent the number of rosette leaves for each plant at bolting; gray lines denote the mean of the number of rosette leaves for each genetic background; error bars show standard deviations determined for at least 20 plants for each genetic background; lowercase letters show significant differences among genetic backgrounds, as determined by Post-hoc Tukey’s HSD test.

To further examine whether the up-regulation of AGL17 contributes to the early flowering phenotype of the clf-29 mutant, we generated clf agl17 double mutants and looked at their flowering time. A T-DNA mutant line SALK_024428, hereafter named agl17-3, was ordered from the Arabidopsis Biological Resource Center. We identified homozygous mutants and examined the expression of AGL17 in WT and agl17-3 by qRT-PCR. Compared to WT, we could barely detect the AGL17 transcript in agl17-3 (Figure 1e), indicating that agl17-3 is a null mutant. We then generated the clf-29 agl17-3 double mutants by genetic crossing. From the F3 progeny, homozygous clf-29 agl17-3 double mutants were identified by PCR-based genotyping. Then, WT, clf-29, agl17-3, and clf-29 agl17-3 were grown in soil side by side under long-day (LD) conditions to estimate their flowering time. Interestingly, we found that clf-29 agl17-3 flowered earlier than WT, but later than clf-29 (Figure 1f). To further confirm that the clf-29 agl17-3 mutant plants flower earlier than the WT but later than the clf-29 plants, we documented the number of rosette leaves at the time of bolting for clf-29 agl17-3 and compared to WT and clf-29. Under LD conditions, clf-29 agl17-3 produced on average 10 rosette leaves, at the time of bolting, which was significantly earlier than WT but later than clf-29 with on average 12 and 8 rosette leaves, respectively (Figure 1g). These data suggest that AGL17 up-regulation is partially responsible for the early flowering phenotype of the clf-29 mutant; or in other words, AGL17 is a significant downstream target of CLF in flowering time control.

Increasing evidence has shown that the floral transition requires PcG-mediated silencing of floral repressors and de-repression of floral activators. Through a transcriptomic approach, we recently found that the floral activators such as AGL19, AGL24, SUPPRESSOR OF OVEREXPRESSION OF CO 1 (SOC1), SEPALLATA 3 (SEP3), and FLOWERING LOCUS T (FT) were all up-regulated in clf-29 compared to WT.¹² Among those activator genes, AGL19 was reported to be repressed by PcG proteins in the absence of cold and the de-repression of AGL19 in clf is partially responsible for the early flowering phenotype of clf mutants.¹⁵ AGL24 and SOC1 have been shown to directly up-regulate each other, thereby providing a positive loop during the floral transition;¹⁶ and both of them have been shown by us to be directly targeted and repressed by CLF,¹² adding the H3K27me3-mediated regulation into the positive-loop of AGL24 and SOC1. It was also uncovered that SEP3 and FT are key for the early flowering phenotype of clf mutants.¹⁷ In addition, CLF also directly targets the floral repressors FLOWERING LOCUS C (FLC) and SHORT VEGETATIVE PHASE (SVP) to regulate flowering.^10,12,18 It is now clear that CLF targets have opposite roles in floral transition; therefore, the clf early flowering phenotype represents the result of both antagonistic and concerted interactions among these factors. This study provides a novel piece of data showing the specific role of the flowering activator gene AGL17 in this complex genetic network.

Funding Statement

Natural Science and Engineering Research Council of Canada [RGPIN/04625-561 2017 to Y.C.]; Agriculture and Agri-Food Canada (to Y.C.); J.S. was supported by a graduate fellowship from the China Scholarship Council.

Disclosure of potential conflict of interest

No potential conflicts of interest were disclosed.